This presentation focuses on the harness architecture that establishes a sustained relationship with your personal AI agent. It is an open-source framework called OpenClaw that orchestrates the memory management and injection pipelines to build trust over time. The system is grown from plain markdown, injected at the start of every turn, and versioned in git. We will cover how identity continuity and memory persist so the model can stand in relation to the user, and the emergent phenomenon that arises from that relationship.
Mei Chen is a research engineer at Renaissance Philanthropy, where she designs and deploys AI systems and harnesses for industrial applications. She spent eight years building machine learning systems, and holds a master's from the University of Waterloo in Systems Design Engineering, researching brain computer interfaces.